1. Purpose
SIRE (Semantic Integrity Retrieval Engine) is an experimental, deterministic
semantic analysis system intended solely for research, evaluation, and
demonstration purposes.
2. Supported Input Format
- English-language datasets only
- CSV files containing text and label columns, or raw English text
- Recommended dataset size: 250–2000 rows
- Maximum file size: 2 MB
- Maximum total text length: 1,000,000 characters
- UTF-8 encoding required
3. Unsupported Usage
- Non-English languages or unsupported formats
- Repetitive, synthetic, prompt-like, or non-dataset text
- Excessive upload attempts or automated load generation
- Probing, inference, or reverse engineering of system logic
- Circumvention of domain construction rules or safety controls
4. System Behaviour
SIRE operates deterministically. Depending on input quality and system
constraints, submitted data may:
- Be rejected during validation
- Abort domain construction
- Produce empty, partial, or trivial output
- Be flagged as out-of-distribution
5. Data Handling
- Uploaded files exist only temporarily for processing
- Temporary data is deleted automatically after processing
- Only hashed identifiers and derived semantic structures may persist
- No raw user text is stored permanently
6. Prohibited Conduct
- Overloading storage, compute, or routing infrastructure
- Creating large numbers of near-duplicate domains
- Reverse engineering SIRE components or encoders
- Bypassing validation or protective mechanisms
- Adversarial testing without explicit authorization
7. Intellectual Property
All SIRE components — including algorithms, tokenization methods, routing logic,
dataset builders, and derived semantic structures — are protected intellectual
property. No rights or licenses are granted beyond permitted use of this system.
8. No Guarantees
SIRE is provided “as is,” without warranties of any kind, including but not
limited to accuracy, performance, or suitability for production use.
9. Acceptance
By accessing or using SIRE, you acknowledge that you have read, understood, and
agree to these terms and commit to responsible use.